Validation of a breast cancer nomogram to predict lymphedema in a Chinese population

Impact of neoadjuvant chemotherapy on breast cancer-related lymphedema after axillary lymph node dissection: a retrospective cohort study

PURPOSE

We aimed to evaluate whether neoadjuvant chemotherapy (NAC) could be a risk factor for breast cancer-related lymphedema (BCRL) associated with axillary lymph node dissection (ALND).

PATIENTS AND METHODS

A total of 596 patients with cT0-4N0-3M0 breast cancer who underwent ALND and chemotherapy were retrospectively analyzed between March 2012 and March 2022. NAC was administered in 188 patients (31.5%), while up-front surgery in 408 (68.5%). Univariate and multivariable Cox regression analyses were performed to determine whether NAC was an independent risk factor for BCRL. With propensity score matching (PSM), the NAC group and up-front surgery group were matched 1:1 by age, body mass index (BMI), molecular subtypes, type of breast surgery, and the number of positive lymph nodes. Kaplan-Meier survival analyses were performed for BCRL between groups before and after PSM. Subgroup analyses were conducted to explore whether NAC differed for BCRL occurrence in people with different characteristics.

RESULTS

At a median follow-up of 36.3 months, 130 patients (21.8%) experienced BCRL [NAC, 50/188 (26.60%) vs. up-front surgery, 80/408 (19.61%); P = 0.030]. Multivariable analysis identified that NAC [hazard ratio, 1.503; 95% CI (1.03, 2.19); P = 0.033] was an independent risk factor for BCRL. In addition, the hormone receptor-negative/human epidermal growth factor receptor 2-negative (HR-/HER2-) subtype, breast-conserving surgery (BCS), and increased positive lymph nodes significantly increased BCRL risk. After PSM, NAC remained a risk factor for BCRL [hazard ratio, 1.896; 95% CI (1.18, 3.04); P = 0.007]. Subgroup analyses showed that NAC had a consistent BCRL risk in most clinical subgroups.

CONCLUSION

NAC receipt has a statistically significant increase in BCRL risk in patients with ALND. These patients should be closely monitored and may benefit from early BCRL intervention.

Comparison of machine learning models to predict the risk of breast cancer-related lymphedema among breast cancer survivors: a cross-sectional study in China

Objective

The aim of this study was to develop and validate a series of breast cancer-related lymphoedema risk prediction models using machine learning algorithms for early identification of high-risk individuals to reduce the incidence of postoperative breast cancer lymphoedema.

Methods

This was a retrospective study conducted from January 2012 to July 2022 in a tertiary oncology hospital. Subsequent to the collection of clinical data, variables with predictive capacity for breast cancer-related lymphoedema (BCRL) were subjected to scrutiny utilizing the Least Absolute Shrinkage and Selection Operator (LASSO) technique. The entire dataset underwent a randomized partition into training and test subsets, adhering to a 7:3 distribution. Nine classification models were developed, and the model performance was evaluated based on accuracy, sensitivity, specificity, recall, precision, F-score, and area under curve (AUC) of the ROC curve. Ultimately, the selection of the optimal model hinged upon the AUC value. Grid search and 10-fold cross-validation was used to determine the best parameter setting for each algorithm.

Results

A total of 670 patients were investigated, of which 469 were in the modeling group and 201 in the validation group. A total of 174 had BCRL (25.97%). The LASSO regression model screened for the 13 features most valuable in predicting BCRL. The range of each metric in the test set for the nine models was, in order: accuracy (0.75-0.84), sensitivity (0.50-0.79), specificity (0.79-0.93), recall (0.50-0.79), precision (0.51-0.70), F score (0.56-0.69), and AUC value (0.71-0.87). Overall, LR achieved the best performance in terms of accuracy (0.81), precision (0.60), sensitivity (0.79), specificity (0.82), recall (0.79), F-score (0.68), and AUC value (0.87) for predicting BCRL.

Conclusion

The study established that the constructed logistic regression (LR) model exhibits a more favorable amalgamation of accuracy, sensitivity, specificity, recall, and AUC value. This configuration adeptly discerns patients who are at an elevated risk of BCRL. Consequently, this precise identification equips nurses with the means to undertake timely and tailored interventions, thus averting the onset of BCRL.

Analysis of the Risk Factors of Breast Cancer-Related Lymphedema and Construction and Evaluation of a Prediction Model

Objective: The occurrence of breast cancer-related lymphedema (BCRL) in postoperative breast cancer survivors is described and the independent risk factors of BCRL are analyzed. A BCRL nomogram prediction model is constructed, and its effectiveness is evaluated to screen out high-risk patients with BCRL. Methods: A univariate analysis was carried out to determine the risk factors possibly related to BCRL, and a logistic regression analysis was utilized to determine the independent risk factors related to BCRL. A BCRL nomogram prediction model was built, and a nomogram was drawn by R software v4.1.0. The area under the curve (AUC) of the receiver operating characteristic (ROC) and the Hosmer-Lemeshow test were used to evaluate the efficacy of the constructed model to assess its clinical application value. Results: The risk factors independently associated with BCRL were body mass index (BMI), handedness on the operation side, no BCRL-related rehabilitation plan, axillary lymph node dissection (ALND), taxane-based chemotherapy, and radiotherapy (all p < 0.05). The BCRL nomogram prediction model was built on this basis, and the results of the efficacy evaluation showed a good fit: AUC = 0.952 (95% confidence interval: 0.930-0.973) for the ROC and χ2 = 6.963, p = 0.540 for the Hosmer-Lemeshow test. Conclusions: The risk factors for BCRL included higher BMI, handedness on the operation side, no BCRL-related rehabilitation plan, ALND, taxane-based chemotherapy, and radiotherapy. In addition, the BCRL nomogram prediction model accurately calculated the risk of possible BCRL among breast cancer survivors and effectively screened for high-risk patients with BCRL. Therefore, this prediction model can provide a basis for rehabilitation physicians and therapists to formulate early and individualized prevention and treatment programs.

Temporal validation of a risk prediction model for breast cancer-related lymphoedema in European population: A retrospective study

AIMS

To perform temporal validation of a risk prediction model for breast cancer-related lymphoedema in the European population.

DESIGN

Temporal validation of a previously developed prediction model using a new retrospective cohort of women who had undergone axillary lymph node dissection between June 2018 and June 2020.

METHODS

We reviewed clinical records to identify women who did and did not develop lymphoedema within 2 years of surgery and to gather data regarding the variables included in the prediction model. The model was calibrated by calculating Spearman's correlation between observed and expected cases. Its accuracy in discriminating between patients who did versus did not develop lymphoedema was assessed by calculating the area under the receiver operating characteristic curve (AUC).

RESULTS

The validation cohort comprised 154 women, 41 of whom developed lymphoedema within 2 years of surgery. The value of Spearman's coefficient indicated a strong correlation between observed and expected cases. Sensitivity of the model was higher than in the derivation cohort, as was the value of the AUC.

CONCLUSION

The model shows a good capacity to discriminate women at risk of lymphoedema and may therefore help in developing improved care pathways for individual patients.

IMPLICATIONS FOR THE PROFESSION AND/OR PATIENT CARE

Identifying risk factors for lymphoedema secondary to breast cancer treatment is vital given its impact on women's physical and emotional well-being.

IMPACT

What problem did the study address? Risk of BCRL. What were the main findings? The prediction model has a good capacity to discriminate women at risk of lymphoedema. Where and on whom will the research have an impact? In clinical practice with women at risk of BCRL.

REPORTING METHOD

STROBE checklist. WHAT DOES THIS PAPER CONTRIBUTE TO THE WIDER GLOBAL CLINICAL COMMUNITY?: It presents a validated risk prediction model for BCRL.

NO PATIENT OR PUBLIC CONTRIBUTION

There was no patient or public contribution in the conduct of this study.

Feasibility and Clinical Utility of Prediction Models for Breast Cancer-Related Lymphedema Incorporating Racial Differences in Disease Incidence

Importance

Breast cancer-related lymphedema (BCRL) is a common complication of axillary lymph node dissection (ALND) but can also develop after sentinel lymph node biopsy (SLNB). Several models have been developed to predict the risk of disease development before and after surgery; however, these models have shortcomings that include the omission of race, inclusion of variables that are not readily available to patients, low sensitivity or specificity, and lack of risk assessment for patients treated with SLNB.

Objective

To create simple and accurate prediction models for BCRL that can be used to estimate preoperative or postoperative risk.

Design, Setting, and Participants

In this prognostic study, women with breast cancer who underwent ALND or SLNB from 1999 to 2020 at Memorial Sloan Kettering Cancer Center and the Mayo Clinic were included. Data were analyzed from September to December 2022.

Main Outcomes and Measures

Diagnosis of lymphedema based on measurements. Two predictive models were formulated via logistic regression: a preoperative model (model 1) and a postoperative model (model 2). Model 1 was externally validated using a cohort of 34 438 patients with an International Classification of Diseases diagnosis of breast cancer.

Results

Of 1882 included patients, all were female, and the mean (SD) age was 55.6 (12.2) years; 80 patients (4.3%) were Asian, 190 (10.1%) were Black, 1558 (82.8%) were White, and 54 (2.9%) were another race (including American Indian and Alaska Native, other race, patient refused to disclose, or unknown). A total of 218 patients (11.6%) were diagnosed with BCRL at a mean (SD) follow-up of 3.9 (1.8) years. The BCRL rate was significantly higher among Black women (42 of 190 [22.1%]) compared with all other races (Asian, 10 of 80 [12.5%]; White, 158 of 1558 [10.1%]; other race, 8 of 54 [14.8%]; P < .001). Model 1 included age, weight, height, race, ALND/SLNB status, any radiation therapy, and any chemotherapy. Model 2 included age, weight, race, ALND/SLNB status, any chemotherapy, and patient-reported arm swelling. Accuracy was 73.0% for model 1 (sensitivity, 76.6%; specificity, 72.5%; area under the receiver operating characteristic curve [AUC], 0.78; 95% CI, 0.75-0.81) at a cutoff of 0.18, and accuracy was 81.1% for model 2 (sensitivity, 78.0%; specificity, 81.5%; AUC, 0.86; 95% CI, 0.83-0.88) at a cutoff of 0.10. Both models demonstrated high AUCs on external (model 1: 0.75; 95% CI, 0.74-0.76) or internal (model 2: 0.82; 95% CI, 0.79-0.85) validation.

Conclusions and Relevance

In this study, preoperative and postoperative prediction models for BCRL were highly accurate and clinically relevant tools comprised of accessible inputs and underscored the effects of racial differences on BCRL risk. The preoperative model identified high-risk patients who require close monitoring or preventative measures. The postoperative model can be used for screening of high-risk patients, thus decreasing the need for frequent clinic visits and arm volume measurements.

Risk prediction models for breast cancer-related lymphedema: A systematic review and meta-analysis

PURPOSE

To review and critically evaluate currently available risk prediction models for breast cancer-related lymphedema (BCRL).

METHODS

PubMed, Embase, CINAHL, Scopus, Web of Science, the Cochrane Library, CNKI, SinoMed, WangFang Data, VIP Database were searched from inception to April 1, 2022, and updated on November 8, 2022. Study selection, data extraction and quality assessment were conducted by two independent reviewers. The Prediction Model Risk of Bias Assessment Tool was used to assess the risk of bias and applicability. Meta-analysis of AUC values of model external validations was performed using Stata 17.0.

RESULTS

Twenty-one studies were included, reporting twenty-two prediction models, with the AUC or C-index ranging from 0.601 to 0.965. Only two models were externally validated, with the pooled AUC of 0.70 (n = 3, 95%CI: 0.67 to 0.74), and 0.80 (n = 3, 95%CI: 0.75 to 0.86), respectively. Most models were developed using classical regression methods, with two studies using machine learning. Predictors most frequently used in included models were radiotherapy, body mass index before surgery, number of lymph nodes dissected, and chemotherapy. All studies were judged as high overall risk of bias and poorly reported.

CONCLUSIONS

Current models for predicting BCRL showed moderate to good predictive performance. However, all models were at high risk of bias and poorly reported, and their performance is probably optimistic. None of these models is suitable for recommendation in clinical practice. Future research should focus on validating, optimizing, or developing new models in well-designed and reported studies, following the methodology guidance and reporting guidelines.

Development and validation of a risk prediction model for breast cancer-related lymphedema in postoperative patients with breast cancer

OBJECTIVE

Breast cancer-related lymphedema (BCRL) is a common post-operative complication in patients with breast cancer. Here, we sought to develop and validate a predictive model of BCRL in Chinese patients with breast cancer.

METHODS

Clinical and demographic data on patients with breast cancer were collected between 2016 and 2021 at a Cancer Hospital in China. A nomogram for predicting the risk of lymphedema in postoperative patients with breast cancer was constructed and verified using R 3.5.2 software. Model performance was evaluated using area under the ROC curve (AUC) and goodness-of-fit statistics, and the model was internally validated.

RESULTS

A total of 1732 postoperative patients with breast cancer, comprising 1212 and 520 patients in the development and validation groups, respectively, were included. Of these 438 (25.39%) developed lymphedema. Significant predictors identified in the predictive model were time since breast cancer surgery, level of lymph node dissection, number of lymph nodes dissected, radiotherapy, and postoperative body mass index. At the 31.9% optimal cut-off the model had AUC values of 0.728 and 0.710 in the development and validation groups, respectively. Calibration plots showed a good match between predicted and observed rates. In decision curve analysis, the net benefit of the model was better between threshold probabilities of 10%-80%.

CONCLUSION

The model has good discrimination and accuracy for lymphedema risk assessment, which can provide a reference for individualized clinical prediction of the risk of BCRL. Multicenter prospective trials are required to verify the predictive value of the model.

Risk factors of unilateral breast cancer-related lymphedema: an updated systematic review and meta-analysis of 84 cohort studies

PURPOSE

To review and update the incidence and risk factors for breast cancer-related lymphedema based on cohort studies.

METHODS

The study was guided by the Joanna Briggs Institute methodology and the Cochrane handbook for systematic reviews. PubMed, EMBASE, CINAHL, Scopus, Web of Science, The Cochrane Library, CNKI, SinoMed, and Wan Fang Database were searched from inception to November 15, 2021. Cohort studies reported adjusted risk factors were selected. PRISMA guideline was followed. Study quality were evaluated using the Newcastle-Ottawa scale. Random-effects models were adopted. The robustness of pooled estimates was validated by meta-regression and subgroup analysis. Lymphedema incidence and adjusted risk factors in the multivariable analyses with hazard / odds ratios and 95% CIs were recorded.

RESULTS

Eighty-four cohort studies involving 58,358 breast cancer patients were included. The pooled incidence of lymphedema was 21.9% (95% CI, 19.8-24.0%). Fourteen factors were identified including ethnicity (black vs. white), higher body mass index, higher weight increase, hypertension, higher cancer stage (III vs. I-II), larger tumor size, mastectomy (vs. breast conservation surgery), axillary lymph nodes dissection, more lymph nodes dissected, higher level of lymph nodes dissection, chemotherapy, radiotherapy, surgery complications, and higher relative volume increase postoperatively. Additionally, breast reconstruction surgery, and adequate finance were found to play a protective role. However, other variables such as age, number of positive lymph nodes, and exercise were not correlated with risk of lymphedema.

CONCLUSION

Treatment-related factors still leading the development of breast cancer-related lymphedema. Other factors such as postoperative weight increase and finance status also play a part. Our findings suggest the need to shift the focus from treatment-related factors to modifiable psycho-social-behavioral factors.

Validation of a nomogram for predicting the risk of lymphedema following contemporary treatment for breast cancer: a large multi-institutional study (KROG 20-05)

PURPOSE

We previously constructed a nomogram for predicting the risk of arm lymphedema following contemporary breast cancer treatment. This nomogram should be validated in patients with different background characteristics before use. Therefore, we aimed to externally validate the nomogram in a large multi-institutional cohort.

METHODS

Overall, 8835 patients who underwent breast cancer surgery during 2007-2017 were identified. Data of variables in the nomogram and arm lymphedema were collected. The nomogram was validated externally using C-index and integrated area under the curve (iAUC) with 1000 bootstrap samples and by calibration plots.

RESULTS

Overall, 1377 patients (15.6%) developed lymphedema. The median time from surgery to lymphedema development was 11.4 months. Lymphedema rates at 2, 3, and 5 years were 11.2%, 13.1%, and 15.6%, respectively. Patients with lymphedema had significantly higher body mass index (median, 24.1 kg/m² vs. 23.4 kg/m²) and a greater number of removed nodes (median, 17 vs. 6) and more frequently underwent taxane-based chemotherapy (85.7% vs. 41.9%), total mastectomy (73.1% vs. 52.1%), conventionally fractionated radiotherapy (71.9% vs. 54.2%), and regional nodal irradiation (70.7% vs 22.4%) than those who did not develop lymphedema (all P < 0.001). The C-index of the nomogram was 0.7887, and iAUC was 0.7628, indicating good predictive accuracy. Calibration plots confirmed that the predicted lymphedema risks were well correlated with the actual lymphedema rates.

CONCLUSION

This nomogram, which was developed using factors related to multimodal breast cancer treatment and was validated in a large multi-institutional cohort, can well predict the risk of breast cancer-related lymphedema.

Who Will Continuously Depend on Compression to Control Persistent or Progressive Breast Cancer-Related Lymphedema Despite 2 Years of Conservative Care?

BACKGROUND

When a patient with breast cancer-related lymphedema (BCRL) depends on continuous compression management, that is, when interstitial fluid accumulation is continuously ongoing, surgical treatment should be considered. Physiologic surgery is considered more effective for early-stage lymphedema. The purpose of this study was to identify predictors of patients with BCRL who will be compression-dependent despite 2 years of conservative care.

METHODS

This study included patients with BCRL who followed up for 2 years. Patients were classified into two groups (compression-dependent vs. compression-free). We identified the proportion of compression-dependent patients and predictors of compression dependence.

RESULTS

Among 208 patients, 125 (60.1%) were classified into the compression-dependent group. Compression dependence was higher in patients with direct radiotherapy to the lymph nodes (LNs), those with five or more LNs resections, and those with BCRL occurring at least 1 year after surgery.

CONCLUSIONS

BCRL patients with direct radiotherapy to the LNs, extensive LN dissection, and delayed onset may be compression-dependent despite 2 years of conservative care. Initially moderate to severe BCRL and a history of cellulitis also seem to be strongly associated with compression dependence. Our results allow for the early prediction of compression-dependent patients who should be considered for physiologic surgery.

A scoring system for predicting the risk of breast cancer-related lymphedema

Objective

We aimed to establish a scoring system to predict the risk of breast cancer-related lymphedema.

Methods

From April 2017 to December 2018, 533 patients who previously underwent surgery for breast cancer were enrolled in this cross-sectional study. Univariate analysis was performed to explore and define the risk factors. A scoring system was then established on the basis of odds ratio values in the regression analysis.

Results

The additive scoring system values ranged from 6 to 22. The receiver operating characteristic (ROC) curve of this scoring system showed a sensitivity and specificity of 83.3% and 57.3%, respectively, to predict the risk of lymphedema at a cut-off of 15.5 points; the area under the curve was 0.736 (95% confidence interval: 0.662–0.811), with χ² = 5.134 (P = 0.274) for the Hosmer–Lemeshow test.

Conclusions

The predictive efficiency and accuracy of the scoring system were acceptable, and the system could be used to predict and screen groups at high risk for breast cancer-related lymphedema.

Supervised Machine Learning Predictive Analytics For Triple-Negative Breast Cancer Death Outcomes

Objective

To use machine learning algorithms to predict the death outcomes of patients with triple-negative breast cancer, 5 years after discharge.

Methods

1570 stage I-III breast cancer patients receiving treatment from Sun Yat-sen Memorial Hospital were analyzed. Machine learning was used to predict the death outcomes of patients with triple-negative breast cancer, 5 years after discharge.

Results

The results showed that platelets, LMR (lymphocyte-to-monocyte ratio), age, PLR (the platelet-to-lymphocyte ratio) and white blood cell counts accounted for a significant weight in the 5-year prognosis of triple-negative breast cancer patients. The results of model prediction indicated that rankings for accuracy among the training group (from high to low) were forest, gbm, and DecisionTree (0.770335, 0.760766, 0.751994, 0.737640 and 0.734450, respectively). For AUC value (high to low), they were forest, Logistic and DecisionTree (0.896673, 0.895408, 0.776836, 0.722799 and 0.702804, respectively). The highest MSE value for DecisionTree was 0.2656, and the lowest MSE value for forest was 0.2297. In the test group, accuracy rankings (from high to low) were DecisionTree, and GradientBoosting (0.748408, 0.738854, 0.738854, 0.732484 and gbm, respectively). For AUC value (high to low), the rankings were GradientBoosting, gbm, and DecisionTree (0.731595, 0.715438, 0.712767, 0.708348 and 0.691960, respectively). The maximum MSE value for gbm was 0.2707, and the minimum MSE value for DecisionTree was 0.2516.

Conclusion

The machine learning algorithm can predict the death outcomes of patients with triple-negative breast cancer 5 years after discharge. This can be used to estimate individual outcomes for patients with triple-negative breast cancer.